|
|
|
Mill Stand
Monitoring Application |
|
Vibration
analysis of Roll Process Mill Stands has proved to be very
successful in solving Mill Stand crashes at an aluminum production
facility. The concept of using vibration analysis was begun at the
facility using a portable data collector, but the Mill Stand cannot
be monitored on an ongoing basis due to personnel safety concerns.
Problem
Statement
 The
Mill Stand has been experiencing chatter which results in aluminum
roll breakage. Suspected causes of the chatter include work roll
resonance, which appears to be related to both the roll speed and
cooling configuration. An experiment involving several combinations
of cooling configuration and roll speed management was conducted
with a temporarily attached monitoring system. Five junk aluminum
rolls were run through the Mill Stand, taking each to failure
(breakage) of the aluminum roll. Before roll breakage, a “standing
wave” in the roll would be observed, with peak-to-peak amplitude
of about 18-24 inches. Roll breakage causes about 45 minutes to 1
hour of unplanned downtime, in addition to quality control problems
caused by the mill stand chatter condition. The goal of the testing
was to see if vibration monitoring could predict in advance that the
Mill Stand was about to enter a chatter condition and if the system
could give the Mill Operator enough warning so the mill could be
reconfigured to avoid chatter and roll breakage.
Mill Stand Monitoring
System Design Guidelines
 |
|
The
data must be transmitted wirelessly to the Ethernet network
connection of the PC server.
|
|
|
The
Master database, which stores configurations, alarms and
statistics, is maintained on the server (optional).
|
|
|
The
Mill Stand monitoring unit should be capable of continued
operation if disconnected from the network.
|
|
|
All
data from the Mill Stand (spectra, waveforms, and analysis
parameters) to be saved in the Microsoft Accessâ
database.
|
|
|
Full
compliment of vibration analysis capabilities including order
tracking, waveform parameters, spectral bands, etc.
|
|
|
The
Monitoring system must be able to report the data and analysis
results in “real time” to the Mill Operator to avoid roll
breakage.
|
Vibration Analysis
 Due
to Mill Stand access limitations, only six sensors were configured
for this special test. In addition, only three analysis parameters
were defined in the database, for the suspected vibration frequency
at resonance and the spectral and waveform overalls. In order to
determine if the vibration data could be used as a “predictor”
of imminent roll chatter, the “Scenario Trend” function of the
PathFinderä
Software was used. The three step process, outlined below, was used
to find a vibration frequency range that trended up in amplitude
level about 60 to 90 seconds before the Mill Stand entered a chatter
condition.
1.
Scenario Trend is found by right-clicking on the measurement
point level in the database tree.
2.
The analyst is then prompted to enter in a frequency range for the
trend.
3. The trend data is
presented for analysis. Analysis of the trend data shows the
predictive capability of the vibration signal. The original trend
data is shown below at right in blue, and the trend of the new
predictive vibration band is shown in pink. Note that in all four
cases the pink trend precedes the blue trend in vibration amplitude
level, therefore predicting that a chatter condition is about to
occur. By setting band frequency and amplitude limits, the operator
can now observe only one gauge to determine if process changes are
about to drive the Mill Stand into a chatter condition. Problem
solved!
|
|
|
|
|
|
